Weitere ähnliche Inhalte
Ähnlich wie IRJET- Deap Base Detection of Epilepsy and Sleep Paralysis through Computed EEG and Face Expression Analysis (20)
Mehr von IRJET Journal (20)
Kürzlich hochgeladen (20)
IRJET- Deap Base Detection of Epilepsy and Sleep Paralysis through Computed EEG and Face Expression Analysis
- 1. International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 05 Issue: 04 | Apr-2018 www.irjet.net p-ISSN: 2395-0072
© 2018, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 105
DEAP BASE DETECTION OF EPILEPSY AND SLEEP PARALYSIS THROUGH
COMPUTED EEG AND FACE EXPRESSION ANALYSIS
Mrs. Buvaneswari B1, Deephika R2, Mythreyee N3, Preethi R4, Shilpa Sundar5
1Associate Professor of B.Tech, Information Technology Department, Panimalar Engineering College,
Tamil Nadu, India.
2,3,4,5 Students of B.Tech, Information Technology Department, Panimalar Engineering College, Tamil Nadu, India.
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Emotion assumes a vital part in our day by day
life and work. Ongoing evaluation and direction of emotion
will enhance individuals' life and improve it. Another case, in
the treatment of patients, particularly those with articulation
issues, the genuine emotion condition of patients will help
specialists to give more suitable medicinal care. As of late,
emotion acknowledgment from EEG has increased mass
consideration. Different highlights and extraction techniques
have been proposed for emotion acknowledgment from EEG
signals, including time space systems, recurrence area
procedures, joint time-recurrence investigation methods, and
different methodologies. Here we proposed a dependable
technique which is proficient in identification of person's
wellbeing like whether they areinfluencedbyEpilepsyorSleep
Paralysis by utilizing (1)Face Expression (2)EEG information.
This plan can identify the state of the distinct individual
utilizing picture classifier and EEG analyzer utilizingversatile
limit recognition strategy
Key Words: Emotion, EEG, Real-time datasets, Face
Expression, Brain Computer Interface
1. INTRODUCTION
Brain-computer interface (BCI) has been one of the most
interesting biomedical engineering research fields for
decades. It provides a promising technology allowing
humans to control external devices by modulating their
brain waves. Most BCI applications have been developed for
non-invasive brain signal processing which is practical to
implement in real-world scenarios. There are plenty of
successful EEG based BCI applications such as word speller
programs [1] and wheelchair controllers [2].Notexclusively
can BCI be utilized to rationally control gadgets, yet in
addition it can be actualized for understanding our
psychological states. Emotion acknowledgment is one of
such applications. Programmed emotion acknowledgment
calculations possibly cross over any barrier amongsthuman
and machine connections. A model of emotion can be
portrayed by two primary measurements calledvalenceand
excitement. The valence is the level of fascination or
abhorrence that an individual feels toward a particular
protest or occasion. It ranges from negative to positive. The
excitement is a physiological and mental condition of being
wakeful or receptive toboosts,extendingfromuninvolved to
dynamic.
2. METHODOLOGY
2.1 System Architecture
We proposed a technique which is productive in recognition
of person's wellbeing like whether they are ordinary or
strange by utilizing two factors fundamentally (1) Face
Expression (2) EEG data. This outline can distinguish the
state of the unique individual utilizing picture classifier and
EEG analyzer utilizing versatile edge recognition technique.
The real emotion state of patients will help doctors to
provide more appropriate medical care. Real-time
assessment and regulation of emotion will improvepeople’s
life and make it better. Here we propose an efficient
algorithm using SURFSIFT method to detect the face
expression of the patient better to tell the patient is affected
by Epilepsy or sleep paralysis. Epilepsy is a kind of
abnormality held in patient at any time during day or night
or at work they keep on showing abnormal face expressions
.The symptoms can be occur throughoutthedayandhold for
so many days. The emotion analysis work enables the
patient to be supervised for a period of time for further
analysis. Using this algorithm, the results are much accurate
and stable. Since, the real time EEG datasets are used we
assure to provide the results effectively and accurately.
2.2 Feature Extraction and Images preprocessing
The info RGB picture is resized to a tallness of 320 pixels.
The resized picture experiences two separate preparing
Fig 1: Block Diagram
- 2. International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 05 Issue: 04 | Apr-2018 www.irjet.net p-ISSN: 2395-0072
© 2018, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 106
pipelines: an immersion based one and shading surface one.
In the first, the picture is initially gamma adjusted and after
that the RGB esteems are changed over to HSV to extricate
the immersion channel. These qualities are naturallylimited
and morphological tasks are connected to tidy up the paired
picture. A moment handling in view of the division
calculation that chips away at both shading and surface
highlights. The info picture of the faceisrecognizedandafter
that it is removed independently. The eyes, nose and the
mouth of the patient’s picture are isolated. The isolated
pictures are put away in the database. These isolated
pictures are utilized for correlation.
2.3 EEG dataset Creation & Group Segmentation
This module comprisesoftechniquesassociatedwithgetting
the EEG informational indexes which coordinates the
individual age gathering and feelings related with the age
and emotional episodes. The EEG informational index is
handled in the MATLAB condition to section it better by
their alpha, beta, Gama, theta ranges. The electrical signs
from the cerebrum are gathered. Thesignsarepreprocessed
and put away. The features are extracted and classified into
Alpha, Beta, Gamma, Theta, and Delta accordingly. The EEG
signals are been generated.
2.4 Classification
This module is utilized to characterizeusingthereal time
EEG dataset and face expressions to detect the patient is
normal or abnormal and happy or sad . These highlights are
additionally utilized for examination motivation behind the
person's passionate conditions and emotional episodes for
mental investigation. It is used to classify the EEG
information with respect to the face expression images to
identify whether the patient is normal or abnormal. If
abnormal then it recognizes whether he is affected by
epilepsy or sleep disorder.
Fig 4: Interest point detection
3. IMPLEMENTATION
The algorithm used here is SURF SIFT algorithm. Both SIFT
and SURF is thus based on a descriptor and a detector.
SIFT is Scale Invariant Feature Transform (SIFT)
Fig 2: Extracted features
Fig 4: EEG signal generation
- 3. International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 05 Issue: 04 | Apr-2018 www.irjet.net p-ISSN: 2395-0072
© 2018, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 107
The SIFT algorithm has 4 basic steps.
[1]Estimate a scale space using the Difference of Gaussian
(DoG). [2]Key points are localized andrefined byeliminating
the low contrast points. [3] Key point orientation in light of
nearby picture angle. [4] Description generator to register
the nearby picture descriptor for each key point in view of
picture inclination extent and introduction.Itrequiresa vast
computational many-sided quality.
The SURF algorithm is Speed up Robust Feature (SURF)
technique which is an approximation of SIFT, performs
faster than SIFT without reducing the qualityofthedetected
points.
SURF approximates the DoG with box filters. Instead of
Gaussian averaging the image, squares are used for
approximation. The SURF uses a BLOB detector which is
based on the Hessian matrix. Fororientationassignmentand
feature description SURF uses the wavelet responses.
The interest points of the megapixel image are detected in
the detection step using the Determinant of Hessian matrix.
In the description step a descriptor is built for the
neighborhood of each point of interest. In Image matching
task the local descriptors from several images are matched.
Exhaustive comparison is performed by computing the
Euclidean distance between all potential matching pairs. A
nearest-neighbor distance-ratio matching criterion is then
used to reduce mismatches.
4. RESULTS AND DISCUSSION
There are several methods to identify the emotion detection
using three emotion detection. But it would not provide the
accurate results and they are not optimistic and unstable.
And also, it does not detect the EEG related diseases.
Table 1: Accuracy level (%)
METHODS ACCURACY
Time-frequency analysis, artificial neural 56.65%
network
Multi wavelet transform, MLPNN 76.89%
Time frequency analysis, KNN 89.87%
HMS analysis, SVM 95.20%
Adaptive threshold analysis, SURF SIFT 98.75%
5. CONCLUSION AND FUTURE ENHANCEMENT
In this project, we are using the EEG results and the face
expressions to detect whether the patient is normal or
abnormal. If abnormal we are obtaining the result as
whether he is affected by Epilepsy or Sleep Disorder. Here,
we are using the Surf Sift algorithm for feature extraction
and to find out the interest points. Using this algorithm, the
results are much accurate and stable. Since, the real time
EEG datasets are used we assure to provide the results
effectively and accurately.
The future enhancement of this project is to find out other
EEG related diseases such as Migraine headache,
Encephalopathy, Sleep Apnoea, Hypersomnia,inan effective
manner.
6. REFERENCES
[1] Z. Wang and A.C. Bovik., “Modern Image Quality
Assessment”. N.Y., Morgan and Claypool Pub. Company,
2006.
[2] R.C. Gonzalez and R.E. Woods, Digital Image Processing,
2nd ed., Prentice-Hall, 2002.
[3] U.K. Sharma and K. Kumawat, “Review of histogram
based image contrast enhancement techniques” in Int.
Journal of Research in Engineering & Technology (IJRET),
Feb 2015, Vol.3, Issue 2, pp.65-76.
[4] W.K. Pratt, Digital Image Processing: PIKS Inside, Third
Edition, New York, John Wiley & Sons, 2001.
[5] E. Yelmanova "Quantitative Evaluation of Contrast for a
Complex Image by its Histogram" in XIII International
Conference TCSET’2016,Lviv-Slavske,Ukraine,2016,p.688-
692.
[6]http://vatechrussia.ru/wpcontent/uploads/2016/05/C
EPH_P ax-3D_006
[7] http://www.angiocalc.com/images/about/Image5.jpg
[8] http://www.neurochirurgie-
genk.be/ziektebeelden/schedel- enhersenen/
hersentumoren/meningeomen/ [9]
https://openi.nlm.nih.gov/detailedresult.php?img=PMC29
4 0258_crn000 1-0015-f01&req=4
[10] Y.T. Kim, “Contrast Enhancement using Brightness
Preserving Bi-Histogramequalization”,inIEEETransactions
on Consumer Electronics, New York, 1997.
[11] R.M. Haralick, Shanmuga K, Dinstein / Textural features
for image classification. IEEE Transactions on Systems Man
and Cybernetics SMC3(6): p.610 – 621, December 1973.
[12] R.A. Vorobel, Logarithmic Image Processing, Kyiv,
Ukraine, Naukova Dumka, 2012.